A new event format, the Zukunfts Bar, attended by Carsten Rother from Heidelberg Collaboratory for Image Processing (HCI) and Jürgen Quittek from NEC Laboratories Europe (NEC), was introduced on Wednesday, March 27. Around 60 guests packed the Heidelberg Technology Park conference center and discussed technology transfer in artificial intelligence (AI) with experts.
Research is ubiquitous in Heidelberg – university and private research facilities dominate the cityscape. But how do research results find their way into application? The Heidelberg Technology Park has been committed to promoting technology transfer in Heidelberg and the entire Rhine-Neckar region for the last 35 years. Transferring scientific achievements and innovations into the public domain is a long-term process, often lasting several years. The scope of the Zukunfts Bar is to identify interfaces between science and business, and promote dialog between the two.
The Zukunfts Bar aims to describe the current state of research and foster discussion between executives and scientists. “The fact that both sides benefit from technology transfer is beyond doubt. In a best-case scenario, research results are verified and applied, and companies market the products, thus generating a basis for public research funding: a perfect value creation cycle,” explains André Domin, Managing Director of the Heidelberg Technology Park and Zukunfts Bar host.
Jürgen Quittek, Managing Director of NEC, and Carsten Rother, Head of Visual Learning Lab Heidelberg, described the current state of research in AI. Let’s say that the computer learns how to analyze images. It should be able to understand the visual data itself. This is achieved through machine learning. At a certain point, the system programs itself and does not necessarily apply human logic. Consequently, it is difficult to understand the computer’s decisions and computational processes. Resolving this issue would accelerate technology transfer, explains Rother: “If you can clearly describe a system, it’s always easier to transfer it to users. And this is what we’re working on at the moment.”
From an industry perspective, the major obstacle to successful AI technology transfer is the lack of qualified personnel: “There are few technically adept engineers who are familiar with AI and understand the difficulty,” explains Quittek. Rother also faces this challenge: “AI is a highly specialized area for which there is an incredibly high demand but only limited human resources.” The two researchers hope that new courses of study at Heidelberg University, such as that in applied computer science, will help redress this situation in the coming years. They both emphasized Heidelberg’s huge potential, particularly for AI in the health sector.
The discussion with the audience revolved around handling Internet data and Heidelberg as a business location. NEC’s Quittek concludes that: “There are many forms of technology transfer in our organization: we create independent spin-offs for very specific research projects or, in the case of medical technology, we introduce artificial intelligence into completely different areas.” Rother adds that technology transfer also occurs between research areas. He is currently investigating the cause of cloud formation with astrophysicists. The HCI conducts parallel research on the fundamentals of image recognition and issues that are relevant to industry. “Industry provides us with critical impetus and, in turn, obtains solutions to its problems … as well as qualified personnel,” says Rother, explaining the interaction between research and industry.
Prof. Carsten Rother studied computer science in Karlsruhe and Stockholm. He was a researcher at Microsoft in Cambridge (Computer Vision Group) form 2003–2013 before accepting a professorship at the Technische Universität Dresden (TUD) from 2014–2017. He has headed the HCI since 2017. His research foci have always been computer vision and machine learning. He is currently involved in spin-offs.
Dr. Jürgen Quittek studied electrical engineering at RWTH Aachen University and received his doctorate from the Hamburg University of Technology (TUHH). He subsequently worked at Berkeley and moved to NEC Europe in 1997, where he has managed the research laboratory and 100 members of staff since 2016. His research focus is artificial neural networks, specifically telecommunications, with network management, data security, energy-efficient communication, and mobile telephony as key areas. He was one of the initiators of the AI research area.
